Cliff Asness, co-founder of AQR Capital Management, made the news recently with the provocative claim that financial markets are getting less efficient. I worked at AQR for 10 years, but long before that I spent nearly two decades as the only mentee the renowned economist Fischer Black ever had. Fischer had a very different view of market efficiency and would, I think, have reached a different conclusion from Cliff’s data.
Market efficiency is more of a slogan than a well-defined concept. There are multiple definitions, and they often give rise to angels on the head of a pin hairsplitting. The Asness paper begins with the statement, “Stock prices should accurately reflect reality,” which is one type of efficiency. Black thought that was impossible.
Consider the simplest reasonable valuation model for stocks, called the “Gordon” model. It says the value of a stock is the dividend it is expected to pay over the next year, divided by the difference between the rate of return investors demand and the expected long-term dividend growth rate. For the S&P 500 Index, the first number is known accurately, the index paid $73 over the last 12 months and is expected to pay $75 over the next 12. The index value of 5,714 suggests investors demand a return 1.31% above the long-term growth rate they expect. For example, if they expect dividends to grow 4% faster than inflation, they want to earn 5.31% above inflation on their S&P 500 index funds.
But hidden in these numbers is the implication that half the present value of the S&P 500 today is represented by cash flows more than 56 years in the future. That is, we have to know the growth rate of the S&P 500 over many decades to value it today. Imagine being in 1968 and trying to guess the growth rate of Meta Platforms Inc. and Alphabet Inc. in 2024.
The Bureau of Economic Analysis estimates the growth rate of gross domestic product. It’s not trying to predict for future centuries, only measure what happened over the previous quarter. It’s often off by 0.2% from its advance estimate a month after quarter end to its second estimate a month later — with more revisions in the future. In chaotic periods, such as the second quarter of 2020 or the fourth quarter of 2008, it can be off by several percent. Remember, this is not the difference between the estimate and the true value, but the difference between two estimates by the same people using the same methodology just with slightly more data.
But a 0.2% difference in the dividend growth rate translates to an 18% error in the value of the S&P 500. Moreover, we know even less about the rate of return investors should demand from stocks, and true equity valuation is far more complicated than the simple Gordon model. Putting it all together, we must confess that no one has any idea what the S&P 500 should sell for.
Fischer went even further. There is no right answer to the question. The S&P 500 value affects the economy, which changes dividend decisions, growth rates and investor preferences. There could be many values for the S&P 500 that would produce consistent cash flows over the next centuries, or no values.
So why have financial markets? Even though no one knows what financial prices should be, we need people to agree on them. We can’t have oil producers planning for one price of gasoline and refiners planning for a different price. Economic activity has to be coordinated, even if that means using a Magic 8 Ball to get prices.
This leads to a more modest idea of market efficiency. Even if prices are guesses, they should move randomly. It should not be easy to beat the market. If the price of oil today is $71 a barrel, but smart people know it will be $72 tomorrow, inconsistent economic decisions will be made. It should be close to an even bet that prices will go up or down, that is, price movements should be close to a random walk.
The Asness paper tests this by looking at the value spread — the ratio of the valuations of expensive stocks to cheap stocks. There are many ways to define expensive and cheap stocks, from simple accounting ratios like price/earnings or price/book, to extensive analyses of the type value investors such as Warren Buffett make. But the value spread for these different measures tend to go up or down together. When value spreads are large, there are big profits to buying cheap stocks and shorting expensive ones. When value spreads are small, that strategy is less attractive.
Most of the time value spreads meander up and down. Occasionally they spike up and then collapse. In extreme cases — 1997-2001, 2006-2009 and 2017-now — we call them bubbles. It’s not just that stocks or some assets soar in price beyond reasonable valuations, it’s that the popular stocks or assets soar well beyond the unpopular ones. Asness argues that this is happening more often than in the past, and that the up-and-down cycles are lasting longer.
There can be little doubt that he is right, and it is surprising that progress in financial theory, far more data delivered faster, more processing power, greater market liquidity and more sophisticated strategies seem to be exacerbating bubbles rather than suppressing them. The paper discusses some possible reasons.
But I think Fischer would have called this an increase in market efficiency, not a decline. We have no reason to think bubble valuations are any worse than normal valuations, or for that matter the valuations in the depth of a panic. What matters is how easy it is to beat the index. One way it can be hard to beat the index is if every asset price follows a perfect random walk with the same risk-adjusted expected return. But another way it can be hard is if the value spread — and other deviations from models — can be big and long-lasting. However big they get, they can get bigger and bankrupt you before you can cash in on the eventual reversion. You can have to stand losses for many years and, as John Maynard Keynes said, the market can remain irrational longer than you can remain solvent.
I think of financial markets as evolved entities that have to protect themselves from clever traders. As traders have gotten more sophisticated, with better and faster data, more theory and bigger computers, markets have responded by killing them with bubbles. A market that fails to evolve dies because clever traders subtract too much profit to make the markets useful for everyone else.
A message from Advisor Perspectives and VettaFi: To learn more about this and other topics, check out some of our webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Aaron Brown
